Indie Business
Tracking Magnet Performance Without Vanity Metrics
Measure claims, reads, and next-purchase behavior you can actually observe. Ignore vanity screenshots.
A magnet is working when the right readers can get the file, some of them finish it, and a few of them take the next step you actually offered. You can observe pieces of that without a dashboard full of borrowed numbers. You cannot observe "brand awareness." You should not pretend a screenshot of someone else's list is a target.
This page is a notebook you can keep in a simple document. It assumes you will not invent a conversion rate, a benchmark, or a promise about sales. If a tool offers those, ignore the decoration and keep the counts you can explain to a friend.
What are you trying to learn, in one question?
Pick the question before you pick the metric. Otherwise you will collect everything and conclude nothing.
Useful questions:
- Are strangers able to claim the file and open it?
- Did the people who claimed it come for this book, or for the word free?
- Of the people who had a clear next step, did any take it?
- Which page sent them, so you know what to fix if the claims are from the wrong audience?
A bad question is "Is my funnel healthy?" You do not have a universal healthy. You have a file, a page, and a list of humans. The piece itself is designed in How to write a reader magnet that converts. If the story is a brochure, no metric will rehabilitate it. You will only measure the brochure more precisely.
Write the question at the top of the note for the month. At the end of the month, answer it in words. The numbers are evidence for the sentence. They are not the sentence.
Which counts are real?
Claims. A claim is a person who asked for the file and left an address you can see. Count claims per week, and write down what you did that week. A mention in the back of the novel, a single newsletter, a quiet page on your site. If you change three things in one week, you will not know which one moved the count. Change one.
Failed deliveries. Count the replies that say the file would not open, and note the device if they told you. A rising claim count with rising failures is a broken file, not a growth story. Fix the export before you send more people. The checklist is in Formatting EPUB ARC files. It applies to magnets too.
Unsubscribes in the first week. These are people who wanted the file and not the ongoing mail, or people who were surprised by what the mail was. Read the timing against your welcome flow. A spike on the day of email two means email two broke the promise on the landing page. The number matters less than the day it happened.
The next step. If the back matter has one link to book two, you can look at clicks on that link or sales you are willing to attribute because there was no other campaign that week. If you used seven links, you chose not to learn. Take links away until the next step is visible.
Replies that mention the story. Do not turn these into a quota. Do read them. "I liked the sister more than the detective" is product information. "Please send me more free books, any genre" is audience information. They imply different fixes.
Source, if you truly have it. A link from the novel's back matter is a different source from a link you posted in a chat. Tag those two links differently if your tool makes that easy. If tagging is becoming a project, use two obvious paths and write down which is which. A notebook line that says "March 3, mentioned in the cozy readers chat" is a source. A pie chart you do not trust is not.
That is the list. You can keep it in a text file. You do not need a warehouse.
What should you refuse to track?
Refuse numbers you cannot define.
A "reach" figure on a social post is not a claim. A follower count is not a reader. An open rate is a guess the mailbox provider may hide, sample, or change without telling you. You can glance at opens if your tool shows them, and you should not manage the book by them. A low open rate on a letter you sent to people who joined for a different genre is a targeting problem, which belongs to When to kill a newsletter swap, not a subject-line emergency.
Refuse benchmarks copied from a thread. "A good magnet converts at 40 percent" is meaningless unless you know the denominator. Percent of what? Page visitors who were already fans? Ad clicks from people who wanted any free file? The denominator is the whole statistic. Without it, the percentage is a mood.
Refuse income forecasts. Claims times a guessed buy rate times a price is a story generator. It will give you a monthly number that feels like a plan. It is arithmetic performed on a wish. If you want money facts, read your store report after the month has happened.
Refuse vanity comparisons with authors who will not show you their traffic source. Their screenshot is an advertisement for the screenshot. You can be glad for them and still not adopt their number as your failure.
How do you look at a month without lying to yourself?
Make a short note with the same headings every time so you are not inventing a new religion in April.
- The question for the month.
- Claims, and the one change you made.
- Delivery failures, and whether you fixed the file.
- What the next step was, and what you saw.
- Two sentences from real replies, copied, not paraphrased into something nicer.
- What you will not change next month.
The last line is the discipline. If you change the magnet, the page, the welcome, the price, and the swap schedule in one month, the next note will be fiction. Hold still long enough to see the file and the page.
Compare months that had similar behavior. A launch week is not the baseline. A holiday is not the baseline. A month where you mailed your own list once and did nothing else is a useful baseline. Write "launch week" in the margin when it is one, so future-you does not treat it as normal.
If the month was quiet, the answer might be that fewer people saw the page. That is distribution. It might be that people saw the page and did not claim. Read the page on a phone and see whether the promise is specific. Vague pages produce vague, small numbers. The fix is the sentence, not a new analytics product.
How do ads and other people's audiences distort the note?
An ad can create claims from people who wanted a freebie and will never open your genre. If you turn on ads, mark those claims as a separate line. Do not average them with the readers who came from the back of the book. You will conclude the magnet "stopped working" when what changed was the room you invited.
The same split applies to group promotions and swaps. A spike from a bundle is not the magnet's organic behavior. Record it as the bundle. If those subscribers leave during the welcome, you learned about the partnership, not about the short story. The partnership filter is in Choosing genres for cross-promo carefully.
Do not buy traffic to a magnet you have not personally claimed on a phone. You will pay to learn that the button was broken. The site and the file come first. Author website basics before you buy ads is the sequence. Tracking does not replace it.
ReaderMagnet can show you claims on a campaign you host here. It will not invent a quality score. List your book if you want that page. Bring your own note. Open beta does not ask for a card until January 1, 2027, and it does not include a magic reader who finishes every file.
What about series book one, if the magnet is the novel?
If the free object is the whole novel, claims are a weaker signal. A download of a free novel is easy to collect and easy to misread. Add the question "Did book two move among people who had book one?" if you can see it at all. Even a rough look at whether book two sold in the weeks after a free run, when you did nothing else, is more honest than celebrating the download total.
Be explicit in the note that you cannot see finishes unless the store or the file delivery actually tells you. Most authors cannot see finishes. Do not estimate them. "I hope half of them read it" is not a metric. Leave the cell blank and write "unknown." Unknown is a true value. A fabricated completion rate is a lie you will quote later by accident.
The tradeoffs of using the novel as the magnet are in Series starters as magnets. The tracking note should say which object you gave away, so you do not compare a 7,000-word story to a free novel as if they were the same offer.
A notebook that stays small enough to keep
Once a month, fill the six lines. Once a quarter, read three notes in a row and write what stayed true. If the true thing is "the file fails on one kind of phone," fix the file. If the true thing is "chat promos bring claims and no replies about the story," stop the chat promo. If the true thing is "the back-of-book link brings a few people who mention the sister character," keep the link and leave the rest alone.
You can use the magnet ideas tool when you are choosing the next piece, not as a scoreboard. Ideas are not performance. Performance is the claim, the failure, the next step, and the reply, written down without a costume.
The work is to stay honest about what you saw. A small note you believe will beat a dashboard you perform for. Keep the note. Change one thing. Write the next book.
How do you write the note when the month is messy?
Some months are not a clean experiment. You got sick. A retailer changed a category. A swap you had forgotten fired on a Tuesday. The note should record the mess instead of forcing a lesson.
Write what happened in the order it happened. "Claims were quiet until the 18th, when the old swap sent. The replies that week asked for the other author's genre. I did not change the magnet." That is a complete finding. It tells you the magnet was not the variable. It tells you the leftover automation was. Turn the automation off, and do not "relaunch" the magnet to compensate. Compensation is how a messy month becomes two messy months.
If you lost the count because a tool reset, say the count is missing. Do not backfill it from memory. A blank week is more useful than a number you invented to keep the column pretty. The point of the notebook is that you can believe it later. A believed blank beats a fictional total.
When you are tempted to add a new metric because the month felt bad, wait until the next note. Add a metric only when a real decision required it and the current lines could not answer. Most bad months are a file problem, a promise problem, or a traffic source problem. You already have lines for those. Fill them in sentences. Then go back to the draft. The notebook is a fence against superstition, not a second job.
What does a useful comparison look like across two magnets?
When you retire a magnet and publish another, keep the notes in the same shape so you can see the change without performing a study. You are not running a laboratory. You are trying not to fool yourself.
Compare the same kind of week. A week when both magnets were mentioned once in the back of the same novel is a fairer look than a week one magnet spent in a group promo. Write the source beside the claims. If the second magnet draws fewer claims and the replies are more specific to the book, that can be a better magnet. Specific replies mean the promise matched. A larger claim count from people who ask for "any free book" is a worse magnet even though the column is taller.
Do not average the two. An average hides the source. Keep them in separate notes and read the sentences. If you cannot decide, wait another quiet month rather than rebuilding both files. Indecision is cheaper than a third version you export on a Sunday because the numbers felt thin. Thin and honest is a result. It often means the book needs readers from the pages you already control, not a new lead magnet to distract you from the novel.